PeopleTracker

PeopleTracker is a detection software that processes live camera or pre-recorded video feeds rapidly and effectively. Upon detection of a person, the software triggers a webhook, similar to Snapshot Cloud, in real-time, thus ensuring swift and efficient people counting metrics.

System Requirements

For PeopleTracker, here are some benchmarks from actual system configurations. You may reference these results in choosing the specification of the system to run PeopleTracker.

a. For a 60 second video:

• Passmark score of 1000, took 433.19 seconds
• Passmark score of 2000, took 368.08 seconds
• Passmark score of 3500, took 250.84 seconds
• Passmark score of 5000, took 130.15 seconds

b. For a 300 second video:

• Passmark score of 1000, took 2096.45 seconds
• Passmark score of 2000, took 1780.92 seconds
• Passmark score of 3500, took 1206.08 seconds
• Passmark score of 5000, took 781.15 seconds

c. For a 600 second video:

• Passmark score of 1000, took 3759.71 seconds
• Passmark score of 2000, took 3188.77 seconds
• Passmark score of 3500, took 2161.32 seconds
• Passmark score of 5000, took 1432.2 seconds

Note: The tests were done using a 720p video at 30 FPS with sample rate set to 1.

For live streaming, here are some recommendations based on actual tests.

• If streaming at 720p and 24 FPS, if the machine has a PassMark score of 2000, please use sample rate of 5.
• If streaming at 720p and 30 FPS, if the machine has a PassMark score of 4500, please use sample rate of 1.

Installation

Abbreviated installation steps below. For a detailed, step-by-step guide, see the Install Docker and SDK on Windows and the associated FAQ for help.

1. Install Docker on your local machine.
2. Get our On-Premise image. In a terminal run docker pull platerecognizer/people-tracker
3. Install and run the docker container:

danger

For the commands below, make sure to change XXXXX to the License Key and YYYYY to your Plate Recognizer Token. Get your Token. Use On-Premise Licenses.

docker run \
  -v /opt/people:/people-data \
  -e LICENSE_KEY=XXXXX \
  -e TOKEN=YYYYY \
  -t platerecognizer/people-tracker

How It Works

The PeopleTracker operates by assigning unique track IDs to detected individuals, providing an efficient means of counting and tracking people within a monitored environment. The track IDs will grow larger, however, users are not guaranteed to receive the ids sequentially (e.g., 1, 2, 3, 4). This happens in situations where a tracked object is unconfirmed as a person; in such cases, the track ID for that particular object will be skipped.

warning

If the same person re-enters the monitored area, they will be assigned another track ID.

danger

If the container running the PeopleTracker is restarted, the counter will reset.

When a tracked object is confirmed as a person within the video stream, PeopleTracker will send a real-time notification via webhook. Read more about it here.